Network analysis method for correlation between sudden sensorineural hearing loss symptom and psychological trouble

Abstract

The invention discloses a network analysis method for correlation between sudden sensorineural hearing loss symptoms and psychological trouble, and belongs to the technical field of disease information processing. Sudden sensorineural hearing loss is hearing loss which attacks rapidly and is unknown in reason, and is often accompanied by symptoms such as tinnitus, dizziness and nausea. Patients often experience psychological troubles, including anxiety and depression, which may aggravate their clinical symptoms. The invention aims to clarify the connections by using network analysis, that is to say, the interaction between the symptoms of sudden sensorineural hearing loss (SSNHL) and a psychological component table is modeled by using the network analysis, and a core node and a connection mode thereof are identified; the key indexes comprise node strength, centrality and edge weight, and are used for quantifying the association between the symptoms and the psychological component table, so that a more personalized and effective treatment strategy can be formulated. Meanwhile, the application of network analysis can deeply understand the polyversity of SSNHL, so that a more personalized and effective treatment strategy can be formulated.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of disease information processing, and particularly relates to a network analysis method for the correlation between sudden sensorineural hearing loss symptoms and psychological distress. BACKGROUND

[0002] Sudden sensorineural hearing loss (SSNHL) is an otologic emergency of unknown etiology, characterized by a ≥30 dB drop in hearing at least three consecutive frequencies within 72 h. This rapid onset of hearing impairment is often accompanied by vestibular symptoms, including nausea, vomiting, dizziness, and persistent tinnitus. The sudden onset of this disease not only impairs hearing function, but also brings significant psychological and social burden, affecting communication ability, professional performance, and overall quality of life. SSNHL is not only a physical disease, but also can have far-reaching emotional consequences, including anxiety and depression caused by the suddenness and often inexplicable nature of hearing loss. Therefore, timely intervention in cases of sudden sensorineural hearing loss is crucial for optimizing rehabilitation outcomes and improving the quality of life of patients.

[0003] SSNHL patients show a high susceptibility to anxiety disorders and depression, with a prevalence rate higher than the general population benchmark. The three diagnostic elements of sudden hearing loss, persistent tinnitus, and vestibular dysfunction create a pathological environment for psychological distress, characterized by acute fear reactions, perceived helplessness, and anticipatory anxiety about prognosis. In addition, long-term auditory deprivation and social interaction disorders can exacerbate depressive symptoms, and persistent tinnitus and dizziness can also contribute to the occurrence of depressive symptoms.

[0004] Comprehensive psychological assessment is an important part of SSNHL management, and anxiety and depression screening is a key prognostic indicator. The Hospital Anxiety and Depression Scale (HADS) is a validated psychological measurement tool that has been widely used to quantify the severity of emotional disorders in patients with physical diseases. When applied to sudden sensorineural hearing loss (SSNHL), HADS can provide valuable insights into the psychological impact of the disease on patients. Recent studies have emphasized the importance of using HADS in assessing and managing anxiety and depression in SSNHL patients. HADS provides a structured approach to addressing the psychological aspects of the disease, and the scale consists of two subscales. By quantifying the psychological distress experienced by patients, HADS scores provide valuable insights into the impact of emotional state on overall treatment effectiveness and quality of life for SSNHL patients.

[0005] In recent years, researchers have proposed that network models can comprehensively describe psychological syndromes. According to network theory, mental disorders can be conceptualized as a complex system composed of interconnected symptoms that reinforce, strengthen, and relate to each other. Network analysis has been applied in the medical field, highlighting highly correlated symptom clusters while using numerical rating scales, thus quantifying the relative importance of individual symptoms. For example, it has been used to describe symptoms and disease progression in schizophrenia, alcohol use disorder, somatic symptom disorder, and panic disorder, as well as depressive and anxiety-related disorders. To describe the association between psychological states and SSNHL, many researchers have assessed the mental health of SSNHL patients. However, none of these studies have described the network structure of HADS and symptoms in SSNHL patients. Network analysis can provide unique insights into symptoms and HADS, thus providing some suggestions for doctors.

[0006] The present invention incorporates network analysis into the study of sudden sensorineural hearing loss (SSNHL) symptoms and psychological distress, revealing the complex interactions between physical manifestations and emotional well-being in SSNHL patients. The main objective of the present invention is to systematically map the dynamic relationship between SSNHL-related symptoms and psychological scales (anxiety and depression) using network modeling, thus identifying key nodes that drive symptom clusters and psychological distress. By combining the Hospital Anxiety and Depression Scale (HADS) with clinical symptoms, the interrelatedness between them is explored. The present invention demonstrates through research that this network analysis provides an overall framework for quantifying symptom centrality, determining the priority of intervention targets, and elucidating how psychological states amplify or maintain physical symptoms. In addition, identifying key nodes such as "stomach butterflies" (A7) related to anxiety and "loss of interest in grooming" (D5) related to depression provides actionable pathways for personalized care. This approach not only improves prognosis accuracy but also lays the foundation for developing evidence-based guidelines that adapt to the biopsychosocial complexity of SSNHL, ultimately improving patient rehabilitation outcomes and quality of life. SUMMARY

[0007] Sudden sensorineural hearing loss (SSNHL) is a rapidly onset, unexplained hearing impairment often accompanied by symptoms such as tinnitus, dizziness, and nausea. SSNHL patients often experience psychological distress, including anxiety and depression, which can exacerbate their clinical symptoms. The present invention aims to elucidate these connections using network analysis, focusing on anxiety and depression, to provide a basis for overall clinical management strategies, thus proposing a network analysis method for the correlation between sudden sensorineural hearing loss symptoms and psychological distress.

[0008] To achieve the above object, the technical scheme adopted by the present invention is as follows:

[0009] A network analysis method for the correlation between symptoms of sudden sensorineural hearing loss and psychological distress, which models the interaction between symptoms and psychological components of sudden sensorineural hearing loss (SSNHL) by network analysis, and identifies core nodes and their connection patterns; key indicators include node strength, centrality and edge weight, which are used to quantify the correlation between symptoms and psychological components, so that more personalized and effective treatment strategies can be developed.

[0010] As a preferred technical solution of the present application, the network analysis method for the correlation between symptoms of sudden sensorineural hearing loss and psychological distress has the following specific steps:

[0011] Step 1, selection of patients with sudden sensorineural hearing loss (SSNHL)

[0012] According to the 2015 edition of "Guidelines for Diagnosis and Treatment of Sudden Sensorineural Hearing Loss", patients diagnosed with sudden sensorineural hearing loss (SSNHL) complete the hospital anxiety and depression scale (HADS) for self-evaluation, thereby obtaining a network analysis data set;

[0013] Step 2, determination and classification of indicators

[0014] Anxiety subscale (HADS-A):

[0015] A1: feeling nervous or agitated; A2: feeling a terrible premonition that something terrible is about to happen; A3: disturbing thoughts fill my mind; A4: can sit and feel relaxed; A5: feel a shivering sense of fear; A6: feel restless, as if you must move; A7: feel a sense of fear, like "butterflies in the stomach";

[0016] Depression subscale (HADS-D):

[0017] D1: still enjoy things you like; D2: can laugh and see the funny side of things; D3: feel happy; D4: feel slow in action; D5: lose interest in your appearance; D6: look forward to things with pleasure; D7: enjoy a good book or radio / TV program;

[0018] Symptoms of sudden sensorineural hearing loss (SSNHL):

[0019] P1: hearing loss; P2: tinnitus; P3: ear fullness; P4: dizziness / dizziness; P5: sleep disorder / insomnia; P6: nausea / vomiting;

[0020] Patient basic information:

[0021] B1: age; B2: gender; B3: years of education;

[0022] Step 3, score statistics

[0023] According to the evaluation indexes determined and classified in step 2, the scores are counted;

[0024] Each symptom of sudden sensorineural hearing loss (SSNHL) is classified as "yes" or "no", represented by 1 and 0 respectively;

[0025] The anxiety subscale (HADS-A) and the depression subscale (HADS-D) each have 7 items, which are scored using a 4-point Likert scale (0-3). The subscale scores are calculated by adding the scores of each item.

[0026] Step 4, network analysis

[0027] Through the enhanced least absolute shrinkage and selection operator (eLASSO) algorithm, the network structure of symptoms and HADS subscales is modeled using qgraph and bootnet software, generating an integrated network topology. Nodes (representing symptoms or HADS subscales) are connected by edges weighted by partial correlation coefficients, with edge thickness proportional to the strength of the association, and color representing directionality (blue: positive correlation, red: negative correlation).

[0028] In network analysis, centrality indicators include strength, closeness, betweenness, and expected influence. Strength represents the sum of the weights of all direct connections between a particular symptom and other nodes in the network. Betweenness quantifies the frequency of a node (e.g., a symptom) acting as a bridge in the shortest paths between other nodes. Closeness measures the proximity of a node to all other nodes in the network, calculated as the inverse of the sum of the shortest path distances to each of the other nodes. These centrality indicators are usually standardized using z-scores for cross-network comparison. Expected influence captures the expected influence of a node on other nodes in the network, taking into account the node's connections, connection strength, and position in the network structure.

[0029] Step 5, stability evaluation of symptom network indicators

[0030] The stability of network indicators is evaluated through case-dropping and bootstrap analysis of network properties.

[0031] The beneficial effects of the present invention mainly include:

[0032] 1. The present invention employs network analysis to model the interaction between symptoms and psychological factors, identifying core nodes and their connection patterns. Key indicators include node strength, centrality, and edge weight, which quantify the association between symptoms and psychological subscales.

[0033] 2. The present invention reveals significant correlations between SSNHL symptoms (e.g., tinnitus, dizziness) and psychological subscales (HADS-anxiety, HADS-depression) through network analysis. Key nodes with high centrality include anxiety, depression, and dizziness, with strong connections between these nodes. Anxiety exhibits the highest node strength, indicating its central role in the symptom-psychological network. Edge weights highlight bidirectional relationships, particularly between anxiety and tinnitus, and between depression and nausea.

[0034] 3. The present invention highlights the critical role of anxiety and depression in the SSNHL symptom network, emphasizing their impact on disease progression and patient well-being. The application of network analysis enables a deeper understanding of the multifaceted nature of SSNHL, allowing for the development of more personalized and effective treatment strategies. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 is a patient dataset selection and exclusion illustration.

[0036] Figure 2 is a network structure diagram of HADS and symptoms. The legend is laid out using the Fruchterman-Reingold algorithm, with colors highlighting statistically-derived clusters.

[0037] A1: Feeling tense or on edge; A2: Feeling a terrible sense of foreboding, as if something terrible is about to happen; A3: Worrying thoughts crowd my mind; A4: I can sit back and feel relaxed; A5: Feeling a shiver of fear; A6: Feeling restless, as if I must move about; A7: Feeling a sense of fear, like "butterflies in the stomach."

[0038] D1: Still enjoy things I used to like; D2: Can laugh and see the funny side of things; D3: Feeling cheerful; D4: Feeling slowed down; D5: Loss of interest in personal appearance; D6: Looking forward with enjoyment; D7: Enjoy a good book or radio / TV programme.

[0039] P1: Hearing loss; P2: Tinnitus; P3: Ear fullness; P4: Dizziness / Vertigo; P5: Sleep disturbance / Insomnia; P6: Nausea / Vomiting.

[0040] B1: Age; B2: Gender; B3: Years of education.

[0041] Figure 3 is a centrality measure diagram of all symptoms in the network.

[0042] Figure 4 Figure 8 is a plot of centrality index stability by case discard bootstrap.

[0043] Figure 5 Figure 9 is a plot of network centrality distribution by clinical group.

[0044] Figure 6 Figure 10 is a plot of patient triage priority distribution by clinical group. DETAILED DESCRIPTION

[0045] The present application is further described in conjunction with the following examples and figures.

[0046] 1. Materials and Methods

[0047] 1.1. Patient Selection and Data Availability

[0048] Clinical data of patients who visited the Department of Otolaryngology of the First Affiliated Hospital of Anhui University of Chinese Medicine from January 2021 to December 2023 were retrospectively collected. This study was approved by the Ethics Committee of the First Affiliated Hospital of Anhui University of Chinese Medicine (2025AH-71).

[0049] All participants were diagnosed with sudden sensorineural hearing loss (SSNHL) according to the 2015 edition of the Guidelines for Diagnosis and Treatment of Sudden Sensorineural Hearing Loss. During hospitalization, patients completed the Hospital Anxiety and Depression Scale (HADS) for self-evaluation, and their medical history and symptoms were assessed by a qualified otolaryngologist.

[0050] Inclusion criteria require: (1) diagnosed as SSNHL according to the 2015 diagnostic guidelines; (2) no mixed diseases, including middle ear lesions, acoustic neuroma, previous head trauma, history of ototoxic drug use or significant noise exposure.

[0051] Exclusion criteria include: (1) transferred to other institutions during treatment, (2) diagnosed with acoustic neuroma or cerebral thrombosis at the same time, (3) incomplete or overlapping data sets (such as Figure 1 shown).

[0052] 1.2. Measurement Indicators

[0053] 1.2.1 Basic Information

[0054] To investigate the demographic characteristics of patients, a new variable containing gender, age, and years of education was calculated. In general, gender is divided into male and female.

[0055] 1.2.2 Symptoms

[0056] Herein, six symptoms were collected for each patient by a specialized ear-nose-throat physician according to the SSNHL guideline (2015). These symptoms were "hearing loss", "tinnitus", "feeling of ear fullness", "dizziness / vertigo", "sleeping disorder / insomnia", and "nausea / vomiting". Each symptom was categorized as "yes" or "no", represented by 1 and 0, respectively.

[0057] 1.2.3 Feature variables (HADS)

[0058] The Hospital Anxiety and Depression Scale (HADS) was developed by British psychiatrists Zigmond and Snaith in 1983 and is a validated self-report tool widely used for screening of anxiety and depressive symptoms in clinical populations. The HADS consists of two separate subscales - the anxiety subscale (HADS-A) and the depression subscale (HADS-D) - with a total of 14 items (7 items per subscale) scored on a 4-point Likert scale (0-3). Items assess the frequency of symptoms over the past week, while to reduce response bias, certain questions are reverse-scored. Subscale scores are calculated by summing the item scores, with total scores ranging from 0 to 21 for both anxiety (HADS-A) and depression (HADS-D), with higher scores indicating more severe symptoms.

[0059] Scores of ≥9 on either subscale have been validated to be optimal for sensitivity and specificity in identifying clinically significant anxiety or depression. Scores of ≤8 are classified as non-clinical. This threshold is useful for efficient triaging of psychological distress in healthcare settings, in line with the tool's intention as a screening aid rather than a diagnostic tool.

[0060] 1.3 Statistical analysis

[0061] After screening, the number of patients meeting the inclusion criteria was 1146. Table 1 shows the patient statistics and descriptive data related to symptoms. Detailed information on the HADS is shown in Table 2.

[0062] Table 1 Patient statistics and descriptive data related to symptoms

[0063]

[0064] Table 2 Detailed information on patient HADS

[0065]

[0066] 2. Analysis

[0067] 2.1 Network analysis

[0068] Network analysis was performed using R software (version 4.3.3). The network structure of symptoms and HADS subscales was modeled using the enhanced least absolute shrinkage and selection operator (eLASSO) algorithm with the qgraph (v1.9.8) and bootnet packages (version 1.6). To optimize sparsity and improve model fit, the penalty coefficient (γ = 0.5) and extended Bayesian information criterion (EBIC) were used to ensure robust selection of node-specific adjacencies.

[0069] The Fruchterman-Reingold algorithm places nodes with stronger, more frequent connections closer together, generating an integrated network topology. Nodes (representing symptoms or HADS subscales) are connected by edges weighted by partial correlation coefficients, with edge thickness proportional to the strength of the association, and color indicating directionality (blue: positive correlation, red: negative correlation).

[0070] Network analysis provides quantitative centrality measures for each node based on the unique configuration of the network. In network analysis, centrality measures include Strength, Closeness, Betweenness, and Expected Influence. Strength represents the sum of the weights of all direct connections between a particular symptom and other nodes in the network. Betweenness quantifies the frequency with which a node (e.g., a symptom) serves as a bridge in the shortest paths between other nodes. Closeness measures the proximity of a node to all other nodes in the network, calculated as the inverse of the sum of the shortest path distances to each of the other nodes. These centrality measures are typically standardized using z-scores to facilitate cross-network comparisons. Additionally, Expected Influence captures the expected influence of a node on other nodes in the network, taking into account the node's connections, connection strengths, and its position in the network structure.

[0071] 2.2 Stability assessment of symptom network indicators

[0072] The stability of network metrics was assessed using case-dropping and bootstrap analyses of network properties. Network stability was operationalized by the resilience of centrality metrics to systematic subsampling, defined as minimal change in node centrality upon progressively excluding data (e.g., up to 75% of cases). This stability was quantified using the correlation stability coefficient (CS-coefficient), with values ​​≥0.5 considered acceptable. The accuracy of edge weights was assessed using a nonparametric bootstrap method (2,500 iterations), generating 95% confidence intervals (CIs) for each association. Bootstrap difference tests further compared the magnitude of edge weights to identify statistically significant variations in network topology.

[0073] 3. Results

[0074] 3.1 Sample feature information

[0075] As shown in Table 1, a total of six symptoms were collected, of which "hearing loss" and "tinnitus" accounted for 99.56% and 94.07%, respectively. Furthermore, 17.71% of the participants had a total score of 9 or higher on the anxiety subscale, and 11.25% had a total score of 9 or higher on the depression subscale. See Tables 1 and 2 for more detailed sample characteristics.

[0076] 3.2 Network structure and centrality measurement analysis

[0077] The network is estimated by the EBICglasso model, and its structure is as follows Figure 2 As shown in Figure 2, 253 edges are retained (average weight Mweight = 0.029). Figure 3 The network centrality metrics, including strength, closeness, betweenness, and expected influence, were displayed. Furthermore, the stability of the network analysis was examined, and it was found to be excellent (i.e., CS coefficient = 0.75). Figure 4 It shows that after discarding 75% of the samples, the network structure will not change significantly compared to the original structure.

[0078] In the anxiety subscale model, node A5 (a feeling of a shiver of fear) was most strongly connected to A7 (a feeling of fear, like "butterflies in the stomach") and A6 (a feeling of restlessness, as if one must move). Node P3 (a feeling of ear fullness) was connected to P4 (dizziness / vertigo), P5 (sleep disturbance / insomnia), and P6 (nausea / vomiting). Node D3 (a feeling of well-being) was connected to D2 (the ability to laugh and see the funny side of things), D5 (loss of interest in one's appearance), and A6 (a feeling of restlessness, as if one must move). Nodes D7 (the ability to enjoy a good book or radio / TV program), D2 (the ability to laugh and see the funny side of things), D4 (a feeling of slowness in one's actions), and A6 (a feeling of restlessness, as if one must move) were connected to A4 (the ability to sit and feel relaxed). Although A4 was on the edge of the network, these nodes were directly connected.

[0079] In terms of strength, node A7 was the most influential. Following A7 were nodes D2, A3, and A6, which were core features of patients with SSNHL. In contrast, other symptoms were on the edge, such as P1, P2, and P3.

[0080] 3.3 Network accuracy and stability

[0081] The network model showed strong reliability, as evidenced by the bootstrap 95% confidence intervals (CIs) of the edge weights obtained from 2,500 resampling iterations, which ensured the stability of the parameter estimates. The difference test based on bootstrapping further confirmed that there were statistically significant differences (α < 0.05) in most of the edge weight comparisons, enhancing the accuracy of the connections between nodes.

[0082] 4. Discussion

[0083] This study used network analysis to explore the interconnections between symptoms and psychological states (particularly anxiety and depression) in patients with sudden sensorineural hearing loss (SSNHL). Using the Hospital Anxiety and Depression Scale (HADS) and clinical data from patients admitted to the Department of Otolaryngology at the First Affiliated Hospital of Anhui University of Chinese Medicine, a network was constructed to elucidate the complex interactions between symptoms related to SSNHL. The results revealed a network structure characterized by significant associations between various symptoms and psychological subscales, providing new insights into this disease.

[0084] The constructed network exhibited strong stability, with a correlation stability coefficient (CS) of 0.75, indicating that 75% of the samples could be excluded without significantly altering the network structure. Centrality analysis identified anxiety-related node A7 ("butterflies in the stomach") as the most influential, followed by D2 ("can laugh and see the funny side of things") and A3 ("worrisome thoughts"). The prominent position of A7 highlights the physiological manifestations of anxiety in SSNHL, which can exacerbate subjective distress and hinder recovery.

[0085] While hearing loss (P1, 99.56%) and tinnitus (P2, 94.07%) were almost universal, less common symptoms such as dizziness / vertigo (P4, 11.08%) and nausea / vomiting (P6, 1.92%) formed distinct clusters with the psychological scales. For example, dizziness / vertigo (P4) was closely linked with ear fullness (P3) and sleep disturbances (P5). Depression symptoms (e.g., D5: "loss of interest in grooming") showed a direct link with anxiety-related restlessness (A6), indicating a bidirectional reinforcement between emotional and physical symptoms.

[0086] The centrality of A7 and D2 highlights actionable intervention targets. For example, cognitive-behavioral strategies targeting somatic anxiety (e.g., mindfulness therapy for gastrointestinal discomfort) can disrupt the feedback loop of A7-A5 ("fearful tremor"). Similarly, enhancing social engagement (through D2 and D3: "feeling happy") can alleviate depression symptoms associated with auditory deprivation.

[0087] Visualization techniques were employed to elucidate the patient stratification patterns derived from network analysis. Figure 5 Boxplot systems were used to compare the distribution characteristics of the three centrality indices (strength centrality, betweenness centrality, closeness centrality) across different clinical subgroups. These plots revealed the essential differences in network characteristics between subgroups: network bridge types exhibited significantly elevated betweenness centrality, while high symptom burden groups showed prominent connection strength centrality. As a supplement, Figure 6 Parallel bar graphs were used to quantitatively present the epidemiological distribution of the cohort in stratified categories. By designing color-coded bar graphs (red for urgent, blue for high-risk, and green for routine), the clinical characteristics of symptom burden distribution can be visually interpreted.

[0088] The aforementioned network analysis represents a significant advancement in understanding the multifaceted nature of SSNHL, surpassing traditional symptom-centered approaches. By elucidating the intricate network between symptoms and psychological states, healthcare professionals can design more personalized and effective treatment strategies that address both the physiological and psychological aspects of the disease.

[0089] 5. Conclusion

[0090] In summary, the study of the present invention demonstrates the feasibility and value of network analysis in studying the symptoms and psychological distress associated with sudden sensorineural hearing loss (SSNHL). The insights gained from this analysis contribute to a deeper understanding of the disease and have the potential to provide more effective and comprehensive care guidance for individuals affected by sudden sensorineural hearing loss.

Claims

1. A network analysis method for the correlation of sudden sensorineural hearing loss symptoms and psychological distress, characterized by, To identify the core nodes and their connection patterns by network analysis modeling the interactions between symptoms and psychological subscales of sudden sensorineural hearing loss (SSNHL), the key indicators including node strength, centrality and edge weight are used to quantify the association between symptoms and psychological subscales, so as to develop more personalized and effective treatment strategies.

2. The network analysis method of claim 1, wherein, The specific steps are as follows: Step 1, selection of sudden sensorineural hearing loss (SSNHL) patients Patients diagnosed as sudden sensorineural hearing loss (SSNHL) according to the 2015 edition of "Guidelines for Diagnosis and Treatment of Sudden Sensorineural Hearing Loss" complete the hospital anxiety and depression scale (HADS) for self-evaluation, thereby obtaining the network analysis dataset; Step 2, determination and classification of indicators Anxiety subscale (HADS-A): A1: feeling nervous or restless; A2: feeling a terrible premonition that something terrible is about to happen; A3: disturbing thoughts fill my mind; A4: I can sit and feel relaxed; A5: feeling a shiver of fear; A6: feeling restless, as if I must move; A7: feeling a sense of fear, like "butterflies in the stomach"; Depression subscale (HADS-D): D1: still enjoy past favorite things; D2: can laugh and see the fun side of things; D3: feel happy; D4: feel slow in action; D5: lose interest in personal appearance; D6: happily look forward to things; D7: enjoy a good book or radio / TV program; Symptoms of sudden sensorineural hearing loss (SSNHL): P1: hearing loss; P2: tinnitus; P3: ear fullness; P4: dizziness / dizziness; P5: sleep disorder / insomnia; P6: nausea / vomiting; Patient basic information: B1: age; B2: gender; B3: years of education; Step 3, score statistics According to the determination and classification of indicators in step 2, the scores are calculated; Each symptom of sudden sensorineural hearing loss (SSNHL) is classified as "yes" or "no", represented by 1 and 0 respectively; Anxiety subscale (HADS-A) and depression subscale (HADS-D) each have 7 items, scored using a 4-point Likert scale (0-3), and subscale scores are calculated by adding the scores of each item; Step 4, network analysis Using the enhanced least absolute shrinkage and selection operator (eLASSO) algorithm, the network structure of symptoms and HADS subscales is modeled using qgraph and bootnet software, thereby generating an integrated network topology; nodes (representing symptoms or HADS subscales) are connected by edges weighted by partial correlation coefficients, with edge thickness proportional to the strength of association, and color representing directionality (blue: positive correlation, red: negative correlation); In network analysis, centrality measures include Strength, Closeness, Betweenness, and Expected Influence; among them, Strength represents the sum of the weights of all direct connections between a particular symptom and other nodes in the network; Betweenness quantifies the frequency of a node (e.g., a symptom) acting as a bridge in the shortest paths between other pairs of nodes; while Closeness measures the proximity of a node to all other nodes in the network, calculated as the inverse of the sum of the shortest path distances from it to every other node; these centrality measures are usually standardized using z-scores for ease of cross-network comparison; additionally, Expected Influence captures the expected influence of a node on other nodes in the network, taking into account the node's connectivity, connection strength, and position in the network structure; Step 5, Stability evaluation of symptom network indicators The stability of network indicators was evaluated through network attribute case-dropping, bootstrap analysis.

3. The network analysis method of claim 2, wherein, The total scores of both the anxiety subscale (HADS-A) and the depression subscale (HADS-D) ranged from 0 to 21, with higher scores indicating more severe symptoms; scores ≥9 on both subscales were optimal for identifying clinically significant anxiety or depression in terms of sensitivity and specificity. Scores ≤8 were classified as non-clinical status.

4. The network analysis method of claim 2, wherein, To optimize sparsity and improve model fitting, a penalty coefficient (γ = 0.5) and extended Bayesian information criterion (EBIC) were used to ensure the robust selection of node-specific adjacency relationships.

Images